Wolbachia Endosymbionts Distort DNA Barcoding in the Parasitoid Wasp Genus Diplazon (Hymenoptera: Ichneumonidae)

Zoological Journal of the Linnean Society, 2016. With 6 ﬁgures

Wolbachia endosymbionts distort DNA barcoding in the parasitoid wasp genus Diplazon (Hymenoptera: Ichneumonidae)

SERAINA KLOPFSTEIN1,2,*, CHRISTIAN KROPF1,2 and HANNES BAUR1,2

1Department of Invertebrates, Naturhistorisches Museum der Burgergemeinde Bern, Bernastrasse 15, CH-3005 Bern, Switzerland 2Division of Community Ecology, Institute of Ecology and Evolution, University of Bern, Baltzerstrasse 7, 3012 Bern, Switzerland

Received 31 July 2015; revised 26 October 2015; accepted for publication 17 November 2015

Molecular species delimitation has the potential to speed up both discovery and description rates for new species. However, several studies have shown a limited performance of the standard DNA barcoding approach which relies on a single mitochondrial gene, cytochrome oxidase 1 (COI). Besides incomplete lineage sorting or a low substitution rate, distortion of mitochondrial inheritance patterns by bacterial endosymbionts could explain the failure of barcoding. Numerous reviews have discussed this phenomenon, but only few empirical examples exist. In the present study, we examine the effect of Wolbachia bacteria on barcoding in the parasitoid wasp genus Diplazon. Although integrative taxonomy recognizes 16 species, COI only recovers up to ten. Adding multivariate morphometrics, genotyping a fast-evolving nuclear gene (ITS2) and screening the Diplazon species for Wolbachia, we show that the failure of DNA barcoding coincides with the presence of the endosymbiont. Two species even share identical COI haplotypes and Wolbachia strains, even though ITS2 suggests that they are not closely related. This is one of very few examples of mitochondrial DNA introgression between well-established insect species, facilitated by Wolbachia. We review similar reports, provide a list of criteria to identify endosymbiont-mediated introgression, and discuss the prevalence and impact of this phenomenon on insect systematics and evolution.

ADDITIONAL KEYWORDS: cytochrome oxidase 1 – hybridization – introgression – species delimitation.

chrome c oxidase subunit 1 (COI), has been advo- INTRODUCTION cated by the Barcode of Life Consortium Molecular data have the potential to greatly speed (www.barcodeoflife.org) (Hebert, Ratnasingham & up the taxonomic workflow and facilitate species dis- deWaard, 2003b; Hebert et al., 2003a). Especially in covery, as demonstrated in numerous studies (e.g., earlier studies, this ‘barcoding’ locus was reported to Blaxter, 2004; Butcher et al., 2012; e.g., Tautz et al., be sufficient to delimit the vast majority of species 2002; Vogler & Monaghan, 2007). Such an increase examined (Hajibabaei et al., 2006; Gomez et al., in the speed at which we describe the biodiversity on 2007; Derycke et al., 2008; Smith et al., 2008; our planet is sorely needed because species are going Schmidt et al., 2015). extinct at unprecedented rates as a result of the Soon, however, reports on inconsistencies between destruction of natural habitats, the introduction of DNA barcoding and established species hypotheses invasive species, and other consequences of human started to accumulate; the prevalence of species activities (Pimm et al., 2014). For animals, the use of non-monophyly and failures of threshold-based a single marker, part of the mitochondrial cyto- delimitation methods were estimated in different studies to range between approximately 10% and 30% (Funk & Omland, 2003; Meier et al., 2006; *Corresponding author. E-mail: [email protected]

Bergsten et al., 2012). The most commonly invoked number of species (Heath et al., 1999; Hurst & Jig- explanations for such discordances are inadequate gins, 2005; Raychoudhury et al., 2009); the success of taxonomy, rapid speciation rates relative to the sub- mtDNA-based species delimitation could thus be stitution rates of the marker, incomplete lineage dependent on the infection history of the species sorting, and hybrid introgression. The first issue, under consideration. inadequate taxonomy, either in the form of misiden- The impact of intracellular endosymbionts can be tifications or poorly supported species hypotheses, is manifold. In the best-case scenario with respect to especially important when unverified data are DNA barcoding, the endosymbionts remain within retrieved from databases, or in poorly studied and one host species (or are transferred between species highly diverse taxa (Collins & Cruickshank, 2012). only infectiously, i.e., not via hybridization; see This issue can only be resolved by careful taxonomic below). By manipulating the reproductive biology of work. The second issue, insufficient variability, can their hosts to maximize their own transmission, as be ameliorated by informed marker choice; variabil- demonstrated, for example, in the a-proteobacterium ity has to be sufficiently high to allow distinguishing Wolbachia, they cause a decrease in the intraspecific between very closely-related species, and some sta- mtDNA variability, in the extreme case causing the tistical approaches for species delimitation even spread of a single haplotype within a whole popula- require some variation at the intraspecific level tion or even species (Turelli, Hoffmann & McKech- (Pons et al., 2006; Zhang et al., 2013). Mitochondrial nie, 1992; Jiggins, 2003; Charlat et al., 2009). They DNA (mtDNA) is advantageous in this respect might thus contribute to the ‘barcoding gap’ (i.e., the because of comparatively high evolutionary rates difference between intraspecific and interspecific and a corresponding high per-base-pair information variation) and thus even improve the performance of content at or around the species boundary (Brown DNA barcoding. Endosymbionts only cause problems et al., 1982; Mindell & Thacker, 1996; Lin & Dan- if an infection is passed on between species through forth, 2004; Mueller, 2006; Simon et al., 2006; Zink a (potentially very rare) hybridization event; under & Barrowclough, 2008); on the other hand, nuclear such a scenario, the bacteria are transferred along- markers (especially introns) often still need to be side a foreign mitochondrium and, if the bacterium established for non-model taxa. The third issue, manages to successfully spread through the new host incomplete lineage sorting, leads to ancestral poly- species through vertical transmission, it will drag morphisms being retained in different, closely- the mtDNA with it. related species, resulting in discordances between The numerous reviews that discuss this mecha- gene trees and species trees (Knowles & Carstens, nism of endosymbiont-mediated mtDNA introgres- 2007; Rosenberg & Tao, 2008; Edwards, 2009; Pru- sion (Johnstone & Hurst, 1996; Jiggins, 2003; fer et al., 2012). Once more, mtDNA is at an advan- Ballard & Rand, 2005; Hurst & Jiggins, 2005; Gal- tage here compared to nuclear DNA because it will tier et al., 2009) draw on very few convincing empiri- attain complete lineage sorting and thus reciprocal cal examples. For a study to provide plausible species monophyly more quickly as a result of the evidence for the role of an endosymbiont in facilitat- four-fold smaller effective population size of the ing mtDNA introgression, it needs to include both mitochondrial compared to the nuclear genome. donor and recipient species and demonstrate a strict By contrast, the last issue (i.e., a higher suscepti- association of both their mtDNA and endosymbiont bility of mtDNA to hybrid introgression) has been strains. To our knowledge, there are currently only discussed as a major drawback of mtDNA (Ballard & six studies that fulfil these requirements (Ballard, Rand, 2005; Hurst & Jiggins, 2005; Bachtrog et al., 2000; Jiggins, 2003; Narita et al., 2006; Whitworth 2006; Rubinoff, Cameron & Will, 2006). After a et al., 2007; Gompert et al., 2008; Raychoudhury hybridization event, the recombining nuclear genes et al., 2009). are continuously eliminated with successive back- In the present study, we provide evidence for crossing, but recombination does not or only very endosymbiont-mediated mtDNA introgression from a rarely occur in mtDNA. Hybridization thus some- group of parasitoid wasps. The genus Diplazon has times leads to the complete introgression of a foreign recently been revised on a mainly morphological mitochondrial genome into another species (Paquin basis, with the discovery of four new species in Eu- & Hedin, 2004; Berthier, Excoffier & Ruedi, 2006; rope (Klopfstein, 2014). Twenty Western Palaearctic Edwards & Bensch, 2009; Galtier et al., 2009; Petit species are currently recognized. We analyzed sev- & Excoffier, 2009; Nicholls et al., 2012). Besides the eral specimens from 15 species from Switzerland and lack of recombination, selective sweeps caused by Sweden and one North American species with DNA hitchhiking with selfish genetic elements such as barcoding and found a very poor recovery rate of the endosymbiotic bacteria have been shown to greatly morphologically defined species. Parasitoid wasps influence the population biology of mtDNA in a might be even more prone to Wolbachia infections

© 2016 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016 WOLBACHIA DISTORTS BARCODING IN DIPLAZON 3 than other insects because of the additional infection specimen did not entirely match the morphological pathway via their arthropod hosts (Cook & Butcher, species concepts in the revision, identifications were 1999; Heath et al., 1999). We thus screened the marked with a ‘cf.’. This was also the case for the wasps for Wolbachia and found that infections specimens of Diplazon tibiatorius with dark hind usually coincide with a failure of DNA barcoding and coxae, which are considered to belong to this species vice versa. To examine the potential role of despite this deviation from the common morphotype Wolbachia in distorting mtDNA inheritance through of D. tibiatorius with orange hind coxae (Klopfstein, hybrid introgression, we compiled a morphometrics 2014). Scientific names and morphological terminol- dataset to support the morphological species ogy are employed sensu Klopfstein (2014). concepts, studied three nuclear and one additional mitochondrial marker for the wasps, and performed MORPHOMETRIC ANALYSIS multilocus strain typing for the endosymbiotic bacteria. The morphological differentiation of Diplazon species is largely based on colour and sculpture of the integument (Klopfstein, 2014). To some extent, such MATERIAL AND METHODS features are subjective and depend on taxonomic expertise. We thus conducted a shape principle com- TAXON SAMPLING AND IDENTIFICATION ponent analysis (PCA) using an independent set of We identified 140 specimens of 15 European species measurements to establish how well the qualitative of Diplazon parasitoid wasps from Switzerland and morphological differentiation is supported by the Sweden (Table 1; see also Supporting information, measurements. We measured 14 characters covering Table S1) using the keys provided in a recent mor- most body parts (see Supporting information, Tables phological revision of the group (Klopfstein, 2014). A S2, S3). The selection corresponds to the measure- North American species, Diplazon bradleyi, was ments used for calculating some of the typically used added together with five outgroup species of the body ratios in Ichneumonidae (Townes, 1969; Klopf- same genus group as Diplazon (Klopfstein, Kropf & stein, 2014). Quicke, 2010; Klopfstein et al., 2011). Because of the We applied the multivariate ratio analysis (MRA) importance of the sculpture of the tergites, mesoscu- of Baur & Leuenberger (2011) to our data. MRA is a tum and mesopleuron for delimiting species in this relatively new approach that is especially suited for genus, we examined all specimens under the same analyzing body measurements in a taxonomy context lighting, a 23-W energy saving lamp. Whenever a because it offers several tools for the analysis of size

Table 1. Summary of taxon and gene sampling

Diplazon species Specimens* Countries† COI ITS2 28S EF1a NADH1

D. albotibialis 2f CH, SE 2 2 2 2 2 D. annulatus 4f, 1 m CH, SF 5 5 D. bradleyi 1f US 1 1 1 1 1 D. deletus 8f CH, SE 8 8 2 1 1 D. ﬂixi 11f, 3 m CH 14 14 2 1 1 D. hyperboreus 2f SE 2 2 1 1 1 D. laetatorius 8f CH, ES, SE, US, ZM 8 8 4 2 2 D. orientalis 3f, 1 m TH 4 4 1 1 1 D. pallicoxa 3f CH, SE 3 3 1 1 1 D. parvus 9f CH, SE 9 6 D. pectoratorius 7f CH, SE, SF 7 7 2 1 1 D. scutatorius 11f CH, SE 11 11 2 2 2 D. tetragonus 19f, 3 m CH, SE 22 21 3 1 1 D. tibiatorius 5f CH, SE 5 5 4 2 2 D. varicoxa 21f, 6 m CH, SE 27 25 3 3 3 D. zetteli 9f, 4 m CH 13 12 1 1 1

*Number of female (f) and male (m) specimens included in the analysis. †Country of origin of the specimens. CH = Switzerland; ES = Spain; SE = Sweden; SF = Finland; TH = Thailand; US = United States of America; ZM = Zambia.

© 2016 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016 4 S. KLOPFSTEIN ET AL. and shape in the multivariate geometrical frame- ENDOSYMBIONT SCREENING AND STRAIN TYPING work (Baur et al., 2014; Schweizer, Hertwig & See- hausen, 2014). Here, we computed a shape PCA and To test whether diplazontine wasps are infected with plotted the first shape PC against isometric size, endosymbionts, we performed a preliminary PCR defined as the geometric mean of all variables (see screening of 32 female specimens from a variety of Supporting information, Fig. S1). Graphic visualiza- species, including eight Diplazon species (Diplazon tion of the correlation of size with shape allowed us deletus, D. flixi, D. laetatorius, D. parvus, D. pectora- to estimate the amount of allometry in the data torius, D. tetragonus, D. varicoxa, and D. zetteli), for (Klingenberg, 1998) (i.e., to judge the impact of an Cardinium hertigii, Rickettsia sp., Spiroplasma indirect size effect on the separation of some taxa). ixodetis, S. poulsonii, and Wolbachia pipientis, using Morphometric analyses were calculated with R, the primers listed in Duron et al. (2008). We version 3.1.2 (R Core Team, 2014), using slightly obtained positive results only in twelve cases for modified R-scripts provided by Baur et al. (2014). Wolbachia, and none for any of the other endosym- Scatterplots were generated with the package ‘gg- bionts. We thus focussed on Wolbachia for further plot20 (Wickham, 2009). analyses within the genus Diplazon. We performed PCR screening on one to ten female specimens per species using primers that amplify a MOLECULAR METHODS fragment of the Wolbachia surface protein gene Genomic DNA was extracted from whole specimens (wsp)(wsp81F and wsp691R; (Braig et al., 1998). preserved in 80% ethanol using the Promega Wizard These primers have been used successfully in the kit for blood and tissue extractions. Vouchers and past to test for Wolbachia infections in Hymenotpera DNA samples are kept at the Natural History (Beukeboom & Pijnacker, 2000) and showed a good Museum in Bern and the Swedish Museum of Natu- performance, especially for supergroup A and B Wol- ral History in Stockholm (Table 1; see also Support- bachia, the two groups previously found in ing information, Table S1). Approximately 700 bp hymenopterans (Simoes~ et al., 2011). As a control for from the 50 end of the mitochondrial COI were ampli- the quality of the extracted DNA, we amplified the fied using the primers LCO and HCO designed by ribosomal 28S gene, which is specific to eukaryotes, Folmer et al. (1994). To obtain approximately 800 bp alongside the Wolbachia screening, using the for- of the nuclear ribosomal RNA (rRNA) internal tran- ward primer designed by Belshaw & Quicke (1997) scribed spacer 2 (ITS2), we used the primers and the reverse primer from Mardulyn & Whitfield designed by Quicke et al. (2006). Three additional (1999) and the PCR conditions described above. Spec- markers, part of the nuclear 28S rRNA (28S), the F2 imens tested positive for Wolbachia in the first round copy of elongation factor 1-a (EF1a) (Klopfstein & were used as positive controls in a second round of Ronquist, 2013), and the mitochondrial NADH 1 screening. gene (ND1), were taken from previous studies (see At least one infected individual per species was Supporting information, Table S1) (Klopfstein et al., then typed using a multilocus sequence typing 2010, 2011). approach relying on five house-keeping genes (Baldo Polymerase chain reactions (PCR) were carried out et al., 2006), complemented by the faster-evolving in 20-ll final volumes using Promega GoTaq Flexi wsp gene; the general primers were used as listed in DNA Polymerase kits. Final volumes contained Baldo et al. (2006) (http://pubmlst.org/wolbachia).

30 pmol of MgCl2, 16 pmol of both primers, 4 pmol of Because multiple infections were detected in several each dNTP, 0.3 U Taq polymerase and 2 ll of geno- species, we used molecular cloning in Escherichia mic DNA. PCR conditions were: 94 °C for 2 min, 35 coli with the Topo Ta Cloning kit (Life Technologies) cycles of 30 s at 94 °C, 30 s at the respective anneal- to separate the different products of the wsp gene ing temperature (51 °C for COI and 49 °C for ITS2), in some specimens (Table 2). Four clones were and 1 min at 72 °C, followed by a final extension at sequenced per specimen. Because multilocus 72 °C for 10 min. PCR products were either purified sequence typing is not possible for multiple infections with the GFXTM DNA and Gel Purification kit without making assumptions about the prevalence of (Amersham Biosciences) or by the purification ser- different allele combinations at the five loci (Arthofer vice of Macrogen Korea. The PCR products were et al., 2011), we only resolved the full sequence typ- sequenced on an ABI 377 automated sequencer using ing for the single infections. Nevertheless, we stud- Big Dye Terminator technology (Applied Biosystems). ied some of the double-peaked sequences of the All new sequences (112 new sequences of COI and multiple infections and found that they were in part 134 of ITS2) have been deposited in GenBank under in accordance with multilocus sequence typing accession numbers KR230498 to KR230743 (see (MLST) strain types already found as single infec- Supporting information, Table S1). tions in other specimens (Table 2); however, as a

© 2016 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016 WOLBACHIA DISTORTS BARCODING IN DIPLAZON 5 result of the highly speculative nature of these asso- (Stamatakis, 2014), with the GTR model and parti- ciations, we only submitted the strain information tioning as above. Clade support was assessed by from single infections to the Wolbachia MLST data- 1000 bootstrap replications. base and to Genbank (accession numbers KR230444 Single-gene trees were also obtained for three to KR230477 for the MLST loci and KR230478 to additional markers (28S, EF1a, ND1) but for a KR230497 for the wsp gene). reduced set of specimens (see Supporting information, Table S1). For the single-gene analyses, we ran MrBayes as specified before but with two indepen- ALIGNMENT AND CALCULATION OF PAIRWISE dent runs only. To obtain a good estimate of the spe- DISTANCES cies tree needed to test for a correlation between The sequences of the protein-coding COI were aligned Wolbachia infections and a failure of barcoding in a with MUSCLE (Edgar, 2004) after translation into phylogenetic contest, a concatenated analysis of all amino acids using MEGA, version 6.06 (Tamura five genes was performed using a single sequence et al., 2013). Alignment was straightforward because sampled randomly for each species. Settings were no indels were detected. In ITS2, the alignment chosen as for the COI and the ITS2 analyses, with posed more problems because of numerous indels of the protein-coding genes partitioned into combined varying length. We thus only aligned the ingroup first and second versus third positions. All analyses taxa and rooted the tree according to previous multi- were run on the University of Bern Linux Cluster gene phylogenetic analyses (Klopfstein et al., 2010, UBELIX. 2011). Alignments can be downloaded from Tree- BASE (http://purl.org/phylo/treebase/phylows/study/ CORRELATION ANALYSIS TB2:S17676). Uncorrected pairwise distances (p distances) were To determine whether the Wolbachia infection status calculated in MEGA with pairwise deletion. Plots showed any correlation with the performance of the were produced in R, version 3.1.2 (R Core Team, mitochondrial COI in species delimitation, we scored 2014). To measure the performance of barcoding, we each species as being infected or not and as being used the threshold approach because it does not recovered in the most sensitive analysis with the require strict reciprocal monophyly of all the species. threshold set at 0.5% uncorrected p distance. Because We tried thresholds of 2%, 1%, and 0.5% uncorrected the Wolbachia infections did not appear to be indepen- p distance. dent from the phylogeny, we had to use a comparative statistical method that corrects for the phylogenetic relationships. We used BayesDiscrete from the Bayes- PHYLOGENETIC ANALYSIS Traits, version 2.0 (Pagel & Meade, 2006) under an Phylogenetic analyses of the COI and ITS2 genes independent and a dependent model of evolution. We were performed under a Bayesian and a maximum run both a ML and a Bayesian approach. For ML, we likelihood (ML) approach. For the Bayesian analyses, used the best-scoring tree found during the ML search we used MrBayes version 3.2.2 (Ronquist et al., on the five-gene dataset in RaxML as a test phy- 2012) with a ‘mixed’ substitution model (integrating logeny. Likelihoods obtained under the two models over the space of possible submodels of the general with 1000 ML attempts per tree were compared by a time-reversible model; Huelsenbeck, Larget & Alfaro, likelihood-ratio test. For the Bayesian approach, we 2004) and gamma-distributed among-site rate varia- evenly sampled 1000 trees from the post-burn-in tion, including a proportion of invariant sites. COI phase of the four independent runs in MrBayes to was run with the combined first and second codon represent the posterior distribution of topologies. The position in one and the third codon position in a sec- dependent and independent models were compared ond partition, ITS2 was unpartitioned. The four using a Bayes factor test, running both in standard independent Markov chain Monte Carlo (MCMC) mode (i.e., assuming a fixed number of four rates runs with one cold and three heated chains were run under the independent and eight under the dependent for 10 000 000 generations and sampled every model) and in the reversible-jump mode (which tries 1000th generation. As a conservative burn-in, we all combinations of equal and zero rates). used half of the samples. Convergence was good as judged from the average standard deviation of split frequencies (ASDSF) for the topology parameter and RESULTS the potential scale reduction factor (PSRF) for the COI BARCODING scalar parameters (ASDSF = 0.0064 for COI and 0.0043 for ITS2, PSRF < 1.002 for all parameters). The sequences of the barcoding portion of COI For the ML analyses, we used RAXML version 8.1 (50 portion of the gene) of 140 specimens of Diplazon

© 2016 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016 Table 2. Wolbachia infection status, barcoding success, and Wolbachia MLST and wsp proﬁles in Diplazon and related species 6 .KLOPFSTEIN S. Wolbachia Individuals Success † ‡ § ¶ Species infection inf/test* barcoding Ind # ST CoxA FbpA FtsZ GatB HpcA wsp HVR1 HVR2 HVR3 HVR4

Diplazon None 0/2 0 albotibialis Diplazon Multi 1/2 0 annulatus Diplazon Multi 1/1 1 ?

bradleyi AL. ET Diplazon Single 1/4 0 706 92 59 17 3 54 68 442 51 55 15 25 deletus Diplazon Single 1/3 0 832 92 59 17 3 54 68 688** 51 269** 15 25 ﬂixi Diplazon None 0/2 1 hyperboreus

06TeLnenSceyo London, of Society Linnean The 2016 laetatorius Diplazon None 0/3 1 orientalis Diplazon None 0/3 1 pallicoxa Diplazon Triple or 3/5 0 625 23/338/? 1/1/17 12/12/9 21/21/? 19/269/18 parvus more Diplazon None 0/3 0 pectoratorius Diplazon None 0/3 1 scutatorius Diplazon Single 1/3 0 843 432** 1 408** 3 88 257** 23 1 12 21 19 tetragonus Diplazon Double or 2/2 0 1B2, 1B4 432** 1 408** 3 88 257** 23/? 1/1 12/12 21/? 19/19 tibiatorius more Diplazon Triple or 10/10 0 825 (1/7) (120/249?) (3/6) (88/8?) (185?) 23/? 1/1 12/56 21/15 19/272 varicoxa more olgclJunlo h ina Society Linnean the of Journal Zoological Diplazon None 0/4 1 zetteli Campocraspedon None 0/1 N/A caudatus Syrphophilus Single 1/1 N/A 514 433** 84 120 200** 234** 257** 689** 11 9 267** 302** asperatus Tymmophorus Single 2/2 N/A 818 52 201** 54 41 75 11 9 15 25 obscuripes Tymmophorus None 0/1 N/A suspiciosus Xestopelta None 0/1 N/A gracilima

*Number of infected/tested individuals. † Success of DNA barcoding: species delimitation unsuccessful (0) or successfull (1) at the most sensitive threshold of 0.5%. ‡ Specimen identifier. § Wolbachia strain type from multilocus sequence typing (MLST), based on the five loci CoxA, FbpA, FtsZ, GatB, and HpcA. 2016 , ¶ Wolbachia surface protein gene (wsp) allele number, and typing of its four hypervariable regions (HVR). **New alleles found in the present study. ?No sequence obtained. WOLBACHIA DISTORTS BARCODING IN DIPLAZON 7 showed surprisingly little divergence between species sequenced a nuclear gene that is known to evolve at of which some were described more than 100 years a relatively fast rate, the rRNA spacer ITS2. The ago and whose status has subsequently remained resulting gene tree (Fig. 5) corresponds well to COI unchallenged in the taxonomic literature. Of the 15 concerning the deeper nodes but shows a very differ- species included with more than one specimen, nine ent pattern for some of the species. Diplazon deletus were not recovered as monophyletic on the majority- and D. flixi do not appear as closely related species rule consensus tree obtained from the Bayesian anal- in ITS2 but, instead, the former clusters with a ysis (Fig. 1). Even more notable, several species North American species, D. bradleyi. There is very shared identical haplotypes: D. deletus with D. flixi, high support for this grouping and, in addition to the D. annulatus with D. tetragonus, and D. parvus with information in the nucleotide sequence, there are D. varicoxa. three indels in the ITS2 sequence that support this Because gene-tree monophyly is not a necessary relationship (of length 1, 2, and 4 bp, respectively). condition for certain species-delimitation methods to These indels are not present in D. flixi or any other work, we used the threshold method as promoted by species grouped with D. deletus in the COI analyses. many barcoding proponents to assess the extent of Diplazon flixi now clusters with specimens of the two discordance between morphologically defined species morphologically very similar species: Diplazon hyper- and those recovered by DNA barcoding. The species boreus and D. zetteli. Neither D. parvus, nor D. vari- recovered when using the threshold method are coxa are recovered by ITS2 as monophyletic, shown in Figure 2. At 2% uncorrected p distance, a although they appear as clearly separated on the value often used for species delimitation in insects current gene tree, with D. parvus now sharing iden- (Hebert et al., 2004), only six species are recovered, tical sequences with D. tibiatorius. This grouping eight at 1%, and ten at the very low value of 0.5%. again is in better accordance with the morphology of Figure 3 shows that the intra- and interspecific the species (Klopfstein, 2014), and suggests that distances overlap broadly in several species pairs. these are good species after all. The species annula- The use of a relative threshold as sometimes pro- tus and tetragonus, however, are not recovered by posed (e.g., ten times the intraspecific variability) ITS2 either but, instead, show a pattern similar to (Hebert et al., 2004) would thus not have improved the COI tree. Finally, several species that are mono- the situation. We here use the distance to the closest phyletic on the COI tree are not supported here (e.g., other species (i.e., the minimum interspecific dis- the aforementioned D. tibiatorius and the species tance) because reporting the average instead of the pair D. scutatorius and D. orientalis, which share smallest interspecific distances exaggerates the bar- identical ITS2 sequences). The two specimens of coding gap (Meier, Zhang & Ali, 2008). D. hyperboreus do not cluster together; one of them showed some deviations from the morphological diag- nosis of the species (see below). MORPHOMETRY The gene trees obtained from the single-gene The morphometric analyses supported most of the analyses of two additional nuclear markers (28S which species hypotheses derived from discrete morphologi- is in close proximity to ITS2 in the genome, and EF1a; cal characters. For all analyses, only the first shape see Supporting information, Fig. S2) show that these PC was informative, which was then plotted markers both evolve too slowly to contain much infor- against isometric size. A scatterplot confined to the mation about the species in question; although the three species pairs of special interest in the present D. albotibialis and D. pectoratorius pair and the group study (D. deletus–D. flixi, D. annulatus–D. tetragonus, around D. laetatorius are recovered, there is little or D. parvus–D. varicoxa) (Fig. 4; see also Supporting no resolution among the questionable species. Not sur- information, Fig. S1) revealed that some of them can- prisingly, the second mitochondrial gene sequenced, not be separated based on quantitative morphology. ND1, largely confirms the picture recovered in the For example, D. annulatus is nested within D. te- COI tree but with less resolution. Diplazon deletus tragonus, and D. parvus within D. varicoxa. How- and D. flixi once more have identical sequences, as ever, D. deletus is clearly distinct from D. flixi by the have D. bradleyi and D. varicoxa (D. parvus and first shape PC (Fig. 4). The two species are also D. annulatus were not sequenced for this locus). entirely overlapping in size; hence, the shape difference cannot be attributed to an allometric size effect. WOLBACHIA INFECTIONS AND CORRELATION ANALYSIS Twenty out of 54 Diplazon individuals or eight of the NUCLEAR ITS2 16 species tested positive for Wolbachia infections, As morphometry did not provide sufficient support (Table 2). To test whether the infections coincided for all the morphological species hypotheses, we with a failure in DNA barcoding, we used a method

Xestopelta gracillima 3C3 Tymmophorus obscuripes 1H2 Campocraspedon caudatus 1C2 Syrphophilus asperatus 514 Tymmophorus suspiciosus 2B11 D. albotibialis 901 0.53/61 D. albotibialis 502 0.84/59 D. pectoratorius 1C7 1.0/99 D. pectoratorius 520 1.0/79 D. pectoratorius 1C6 0.63/38 D. pectoratorius 3H2 D. pectoratorius 4F3 0.86/- D. pectoratorius 815 D. pectoratorius 907 D. laetatorius 1C4 D. laetatorius 1C5 D. laetatorius 3A2 1.0/85 D. laetatorius 3B2 0.99/98 D. orientalis 707 D. laetatorius 4B10 D. orientalis 701 1.0/95 D. laetatorius 4G1 D. orientalis 711 D. laetatorius 712 D. orientalis 704 D. laetatorius 703 D. scutatorius 1A10 D. scutatorius 1A11 1.0/100 D. scutatorius 2A3 D. scutatorius 2E12 D. scutatorius 2E3 D. scutatorius 4B12 1.0/99 1.0/95 D. scutatorius 4D4 D. scutatorius 4E3 0.01 substitutions per site D. scutatorius 4F2 D. pallicoxa 503 D. scutatorius 709 D. pallicoxa 601 D. cf. scutatorius 845 1.0/98 D. pallicoxa 618 D. annulatus 1C1 D. tetragonus 2A9 D. annulatus 0.76/- D. annulatus 4G6 D. annulatus 2D6 D. tetragonus 4F6 D. annulatus 2D4 1.0/- D. annulatus 2D7 D. deletus 1A7 D. tibiatorius 1B2 D. deletus 1A8 1.0/98 D. tibiatorius 1B3 D. flixi 4H1 D. cf. tibiatorius 1B4 D. deletus 2E9 D. cf. tibiatorius 3E1 D. deletus 4H10 D. cf. tibiatorius 3F1 D. flixi 1A1 D. zetteli 1A3 D. deletus 714 D. zetteli 1A4 D. flixi 4D12 D. zetteli 2D2 D. deletus 706 D. tetragonus D. zetteli 2D3 D. flixi 1A2 D. zetteli 2D5 D. flixi 2E5 1.0/98 D. zetteli 2E1 D. flixi 2E6 D. zetteli 2E2 D. deletus 2E10 0.97/63 D. zetteli 3A1 D. flixi 4G11 D. zetteli 3B1 0.65/- D. flixi 4G12 D. zetteli 3B4 D. flixi 4G7 D. zetteli 3C1 D. flixi 4G9 D. zetteli 621 D. deletus 4H12 D. zetteli 622 D. flixi 4H3 D. hyperboreus 903 D. flixi 4H5 1.0/68 D. cf. hypberboreus 619 D. flixi 4H8 D. flixi 832 D. bradleyi 2A5 D. tetragonus 1A12 D. parvus 623 D. tetragonus 1B1 D. varicoxa 620 D. tetragonus 2A10 D. parvus 625 D. tetragonus 2A4 D. deletus D. varicoxa 909 D. tetragonus 2A7 D. varicoxa 848 0.58/49 D. tetragonus 2A8 D. parvus 629 D. tetragonus 3B3 D. varicoxa 616 D. tetragonus 3C2 D. cf. parvus 905 0.62/- D. tetragonus 3E2 D. cf. parvus 913 D. parvus D. tetragonus 3G2 D. varicoxa 1A6 D. tetragonus 4B7 D. varicoxa 2E11 D. tetragonus 4E11 D. varicoxa 2E7 D. tetragonus 4G3 D. varicoxa 2E8 D. tetragonus 809 D. varicoxa 3D1 D. tetragonus 843 D. varicoxa 4D3 D. tetragonus 911 D. varicoxa 521 D. tetragonus 3A3 D. parvus 624 D. tetragonus 3D2 1.0/90 D. varicoxa 630 D. tetragonus 3E3 D. varicoxa 631 D. tetragonus 3F2 D. varicoxa 632 D. varicoxa 825 D. flixi D. varicoxa 846 D. parvus 627 D. varicoxa D. varicoxa 921 D. varicoxa 1A5 D. varicoxa 1A9 D. varicoxa 613 D. varicoxa 614 D. varicoxa 615 D. parvus 844 D. varicoxa 633 D. parvus 626 D. cf. varicoxa 902 D. cf. varicoxa 923 D. varicoxa 2F1 D. cf. varicoxa 617

Figure 1. Bayesian majority-rule consensus tree as retrieved from the barcoding fragment of COI mtDNA. Support values close to the nodes represent Bayesian posterior probabilites and bootstrap support based on 1000 replicates. Inlaid photographs show specimens of some of the unresolved species. Part of the tree was cut at the triangle and moved to the left to ﬁt on a single page.

Threshold le i utat e r lixi hyper bradl par zetteli tibiatorius d f pallicoxa albotibialis pectoratorius o laetatorius sc annulatus tet varicoxa 2% 1 2 3 4 5 6

1% 1 2 3 4 5 6 7 8

0.5% 1 2 3 4 5 6 7 8 9 10

Figure 2. Numbers of species recovered and identity of lumped species as obtained by the threshold method for three different threshold values. Distances are uncorrected pairwise distances in the CO1 barcoding locus.

Intra-specific distances inter-specific distances (to closest species) orientalis albotibialis pectoratorius albotibialis tetragonus pectoratorius tibiatorius orientalis scutatorius deletus annulatus tetragonus annulatus hyperboreus tetragonus flixi deletus deletus varicoxa parvus zetteli varicoxa orientalis parvus laetatorius flixi scutatorius pallicoxa 0.00 0.01 0.02 0.03 0.04 0.05 0.06 flixi zetteli parvus deletus varicoxa pallicoxa orientalis tibiatorius annulatus laetatorius albotibialis tetragonus scutatorius pectoratorius

Figure 3. Intra- and interspecific uncorrected p distances for each species. The minimum interspecific distances are shown (i.e., those to the closest species included in our dataset; species identities are indicated above the respective box- plot). Grey bars indicate species that do not show any barcoding gap (i.e., for which intra- and interspecific distances overlap). correcting for phylogenetic relationships by compar- dard MCMC = 5.14, from reversible-jump MCMC = ing an evolutionary model assuming independent 2.84; Bayes factors are considered significant from a with one based on dependent evolution. Barcoding value of 2 and highly signficiant from a value of 5; was considered successful if the species were recov- Kass & Raftery, 1995]. ered at the most sensitive threshold of 0.5%. Both To examine the extent to which this method was the ML approach and the Bayesian approach sensitive to taxon sampling, we also analyzed the significantly preferred the dependent over the inde- dataset under the assumption that the aberrant pendent model of evolution in both ML model testing specimen of D. hyperboreus, which was retrieved (likelihood ratio test statistic = 10.29, P < 0.0358) apart from the specimen with the typical morphology and Bayesian testing [Bayes factor as 2 9 (difference in the ITS2 tree, actually represents a different in the logarithm of the harmonic mean) from stan- species. This change had a drastic impact on the

Species the USA (Wolbachia MLST database, http:// deletus pubmlst.org/wolbachia; accessed 20 May 2015). Dipla- flixi zon tetragonus has two previously unknown alleles for the genes FbpA and HpcA and thus harbours a new strain. Interestingly, the wsp allele found in this species has previously been found in species as diverse as 0.2 the parasitic wasp Nasonia longicornis (Darling) (Hymenoptera, Pteromalidae), the fruit ﬂy Rhagoletis cerasi (Linnaeus) (Diptera, Tephritidae), and the vine- 0.1 gar ﬂy Leucophenga maculosa (Coquillett) (Diptera, Drosophilidae) (http://pubmlst.org/wolbachia). Dipla- zon tibiatorius carries multiple though closely related 0.0 wsp alleles, although MLST sequence typing only

PC1 (50.9 %) recovered a single strain, which is identical to the one found in D. tetragonus. Finally, the infections of –0.1 D. varicoxa and D. parvus could not be typed by the MLST approach because of multiple infections (triple or more), although one of the cloned wsp sequences

–0.2 corresponds to the same allele (#23) as the infections in D. tetragonus and D. tibiatorius (Table 2). The phy- –0.3 –0.2 –0.1 0.0 0.1 0.2 logeny of the wsp sequences of the single-infected spe- Isometric size cies and the successfully cloned multiple infections (Fig. 6) confirms the strain typing results. Figure 4. Scatterplot of isometric size versus the first shape principle component of the species pair Diplazon deletus–Diplazon flixi. PC, principal component. DISCUSSION

FAILURE OF DNA BARCODING IN DIPLAZON result, which thus has to be considered with caution. With this new dataset, the significance disappeared, We found a rather poor performance of the standard with the likelihood ratio test statistic dropping to DNA barcoding approach in delimiting species in the 5.47 and the P value increasing to 0.24. parasitoid wasp genus Diplazon. Using a 2% sequence divergence threshold as often advocated (Hebert et al., 2004), only six of the 16 species could WOLBACHIA DIVERSITY be recovered, making this approach clearly insuffi- The Wolbachia surface protein gene (wsp)sequences cient in this genus. This result is somewhat in con- indicated single infections in D. deletus, D. flixi, trast to other studies that have reported good D. tetragonus, and the outgroup species Syrphophilus success of barcoding in other groups of parasitoids, asperatus and Tymmophorus obscuripes. Multiple where COI typically recovered many more species infections, as evident from polymorphic peaks and than morphology (Smith et al., 2008; Stigenberg & often also length variation in the wsp sequences, were Ronquist, 2011; Butcher et al., 2012). However, mor- detected for D. annulatus, D. bradleyi, D. parvus, phological examinations were often not very detailed D. tibiatorius,andD. varicoxa. For the latter three, in the past and most studies covered only limited we performed molecular cloning of the wsp gene to geographical regions. Further studies will show assess how many infections were present and whether whether the poor performance in Diplazon is just an they were similar to single infections already detected exception for the group. A simple DNA-based method in the present study (Table 2). for species delimitation would be urgently needed in Multilocus sequence typing of the singly-infected parasitoid wasps, given their enormous and highly species recovered one known and three unknown Wol- understudied diversity (Quicke, 2012). On the other bachia strains (Table 1). Diplazon deletus and D. flixi hand, parasitoids might have extreme population are both infected with the same strain #92 (http:// dynamics because of their high trophic level, show pubmlst.org/wolbachia). The sequences of the fast- unusually biased sex-ratios and even high levels of evolving wsp gene differs only by a single mutation in inbreeding in some groups, and potentially have very the second hypervariable region of the gene, which is fast speciation rates through host switching and further evidence for the very close relationship ecological speciation (Feder & Forbes, 2010; Konig€ between those Wolbachia infections. The same strain et al., 2015). Furthermore, parasitoids might be more has also been found in two lepidopteran species from prone to Wolbachia infections as a result of their

D. albotibialis 502 0.75/51 D. albotibialis 901 D. pectoratorius 1C7 D. laetatorius 3B2 1.0/100 D. pectoratorius 520 D. laetatorius 1C4 D. pectoratorius 4F3 D. laetatorius 1C5 D. pectoratorius 1C6 0.95/76 D. laetatorius 3A2 D. pectoratorius 3H2 D. laetatorius 4B10 0.51/- D. pectoratorius 815 1.0/100 D. laetatorius 4G1 D. pectoratorius 907 D. laetatorius 712 D. laetatorius 703 D. orientalis 711 D. scutatorius 2E3 1.0/100 D. cf. scutatorius 845 D. scutatorius 709 D. scutatorius 1A10 D. scutatorius 1A11 D. scutatorius 2A3 1.0/100 D. scutatorius 2E12 1.0/96 D. orientalis 701 D. scutatorius 4B12 D. scutatorius 4D4 D. scutatorius 4E3 D. scutatorius 4F2 D. orientalis 707 D. orientalis 704 D. pallicoxa 503 1.0/100 D. pallicoxa 601 D. bradleyi 2A5 D. pallicoxa 618 D. deletus 1A7 1.0/100 D. deletus 1A8 1.0/100 D. deletus 2E10 D. deletus 2E9 1.0/89 D. deletus 4H10 D. deletus 4H12 D. deletus 714 D. deletus 706 D. tetragonus 2A8 0.99/67 D. annulatus 1C1 D. annulatus 2D4 D. annulatus 2D6 D. annulatus 2D7 1.0/86 D. annulatus 4G6 D. tetragonus 2A9 D. tetragonus 4E11 1.0/100 D. tetragonus 4F6 0.005 substitutions per site D. tetragonus 1A12 D. tetragonus 1B1 D. tetragonus 2A10 D. tetragonus 2A4 D. tetragonus 2A7 D. tetragonus 3A3 D. varicoxa 1A5 0.76/53 D. tetragonus 3B3 D. cf. varicoxa 617 D. tetragonus 3C2 D. tetragonus 3D2 D. varicoxa 2E11 D. tetragonus 3E2 D. cf. hyperboreus 619 D. tetragonus 3E3 D. varicoxa 2F1 D. tetragonus 3F2 D. tetragonus 3G2 D. tibiatorius 1B2 D. tetragonus 4B7 D. parvus 624 D. tetragonus 4G3 D. parvus 627 D. tetragonus 8 09 D. cf. tibiatorius 1B4 1.0/87 D. tetragonus 8 43 D. parvus 625 D. flixi 1A1 1.0/86 D. parvus 629 D. flixi 1A2 D. tibiatorius 1B3 D. flixi 2E5 1.0/96 D. parvus 623 D. flixi 2E6 D. parvus 626 D. flixi 4D12 D. cf. tibiatorius 3E1 D. hyperboreus 903 cf. D. tibiatorius 3F1 0.55/- D. flixi 4G12 D. varicoxa 1A6 D. flixi 4G7 D. varicoxa 2E7 D. flixi 4G9 D. varicoxa 4D3 D. flixi 4H1 D. varicoxa 521 D. flixi 4H3 D. varicoxa 631 D. flixi 4H5 D. varicoxa 620 D. varicoxa 630 D. flixi 4H8 D. varicoxa 632 D. flixi 832 D. varicoxa 633 D. flixi 4G11 D. zetteli 2E1 1.0/77 D. varicoxa 846 D. varicoxa 848 1.0/92 D. zetteli 622 D. varicoxa 825 D. zetteli 621 D. varicoxa 921 D. zetteli 3C1 D. varicoxa 1A9 D. zetteli 3B1 D. varicoxa 2E8 D. zetteli 3A1 D. varicoxa 3D1 D. zetteli 2E2 D. varicoxa 613 D. zetteli 2D5 D. varicoxa 614 D. varicoxa 615 0.94/61 D. zetteli 2D3 D. varicoxa 616 D. zetteli 2D2 D. varicoxa 909 D. zetteli 1A4 D. zetteli 1A3

Figure 5. Bayesian majority-rule consensus tree as retrieved from the internal transcribed spacer 2 rRNA. The Dipla- zon species are shown in different colours. Support values close to the nodes represent Bayesian posterior probabilites and the bootstrap support based on 1000 replicates. The branch leading to Diplazon albotibialis and Diplazon pectoratorius has been shortened to ﬁt on a single page.

Syrphophilus asperatus 514 1.0 Tymmophorus obscuripes 1H2 0.75 Diplazon parvus 624 Clone 1 706 1.0 Diplazon deletus Diplazon flixi 1A2 0.98 Diplazon flixi 832 1.0 Diplazon varicoxa 825 Clone 1 Diplazon parvus 624 Clone 2 1.0 Diplazon parvus 624 Clone 3 Diplazon tetragonus 843 1.0 Diplazon_tibiatorius 1B2 Clone 1 Diplazon tibiatorius 1B2 Clone 2 1.0 Diplazon cf. tibiatorius 1B4 Clone 1 Diplazon cf. tibiatorius 1B4 Clone 2 Diplazon varicoxa 825 Clone 2 0.01 substitutions per site Diplazon varicoxa 825 Clone 3

Figure 6. Bayesian majority-rule consensus tree of the Wolbachia surface protein (wsp) sequences of isolates from nine species of Diplazontinae parasitic wasps. Strains separated by molecular cloning in Escherichia coli were given arbitrary numbers. Values next to nodes represent Bayesian posterior probabilities. intimate relationships with their hosts that might barcoding. However, this result should be treated act as a potent transmission pathway (Cook & with caution because it was very sensitive to the spe- Butcher, 1999; Vavre et al., 1999). All of these fac- cies hypotheses. This was exemplified by our exercise tors might complicate the population biology of their of assuming species status for the aberrant individ- mtDNA and thus impede DNA barcoding. ual of D. hyperboreus, which was sufficient to annihi- In Diplazon, only a combination of the two mark- late our result. The test is probably not very stable ers COI and ITS2 recovered most of the morpholog- because of the very small number of taxa sampled ically defined species; in some cases, for example, (eight infected versus eight non-infected putative the recently described D. parvus, species status is species) and very short branch-lengths in the crown only supported by the combined information because group of the tree. Furthermore, the infection status this species was polyphyletic in both markers but of a species might have changed recently or not be with respect to two different species (Figs 1,4). detected correctly in our few-specimen assay. Given the limited use of COI and ITS2 as markers A recent critique of phylogenetic comparative for species delimitation, the establishment of addi- methods in general highlighted a basic shortcoming tional markers is necessary. A recent bioinformatics in that they often even retrieve a significant correla- approach using comparative genomics (Hartig et al., tion if the character histories involve only a single 2012) has already established a plethora of candi- origin (Maddison & FitzJohn, 2015). Currently, there date loci for the order Hymenoptera, some of which is no way to resolve this issue except for the careful contain introns that might provide sufficient vari- interpretation of such a result. In our case, a single ability to resolve questions at the species level, and origin of both characters can be excluded from the could be analyzed in multispecies coalescent phylogenetic distribution, although the species with approaches that make use of the information inher- both a failure of barcoding and Wolbachia infections ent in independently segregating markers to iden- are certainly concentrated among the crown group of tify reproductively isolated units (Yang & Rannala, the tree. The failure of barcoding might be tightly 2010). In any case, the failure of a single-marker linked to lowered evolutionary rates or rapid specia- identification system in this genus suggests that tion rates, and a phylogenetic component is likely for caution is necessary when using DNA taxonomy in both. By contrast, the observed Wolbachia infections parasitic wasps. certainly do not go back to a single infection in an ancestral species because some strains are very divergent and several species harbour multiple strains. CORRELATION BETWEEN WOLBACHIA INFECTIONS AND Furthermore, the retention of Wolbachia over such A FAILURE OF BARCODING time scales and that numbers of species boundaries is We found a significant correlation between the highly unlikely for this endosymbiont, which nor- Wolbachia infection status and the failure of mally shows much faster infection dynamics (Werren,

Baldo & Clark, 2008; Raychoudhury et al., 2009). cordance between nuclear and mitochondrial gene Even when accepting the correlation as true, there trees in the vinegar fly genus Drosophila (Diptera, might still be other causal explanations than Drosophilidae) with an introgression event from endosymbiont-mediated hybrid introgression (e.g., a Drosphila simulans Sturtevant to Drosphila mauri- role of Wolbachia in increasing speciation rates, tiana Tsacas and David, and such a transfer could which would at the same time decrease the success of even be repeated experimentally (Aubert & Solignac, DNA barcoding) (Werren et al., 2008; Raychoudhury 1990). Jiggins (2003) found a shared mtDNA haplo- et al., 2009). We thus need additional evidence to type in those individuals of two species of Acraea support the introgression scenario. butterflies (Lepidoptera, Nymphalidae) that were infected with the same Wolbachia strain but not in the uninfected individuals, and such a pattern is best A STRONG CASE FOR WOLBACHIA-MEDIATED explained by endosymbiont-mediated hybrid intro- MTDNA TRANSFER gression. Narita et al. (2006) recovered the same Additional evidence for Wolbachia-mediated mtDNA pattern in two recently discovered sibling species of introgression comes from several sources. Under a the butterfly genus Eurema (Lepidoptera, Pieridae). scenario of hybrid introgression, we expect the follow- Whitworth et al. (2007) examined twelve species of ing patterns: (1) very low mtDNA diversity indicative the blowfly genus Protocalliphora (Diptera, Cal- of a recent selective sweep; (2) mtDNA haplotypes liphoridae) and found that four species shared COI that are much more similar than likely given the spe- haplotypes and Wolbachia strains as judged from the cies relationships; (3) identical or at least very similar wsp gene, whereas AFLP markers suggested that Wolbachia infections; and (4) a likely opportunity for these species were not closely related. Similar results hybridization (Hurst & Jiggins, 2005). were obtained by Gompert et al. (2008) in Lycaeides For the species pair D. deletus–D. flixi, without any butterflies (Lepidoptera, Lycaenidae). Finally, Ray- doubt two good biological species that were also choudhury et al. (2009) examined the Wolbachia clearly distinct in our morphometric analysis (Fig. 4), infections in the parasitoid wasp genus Nasonia all four of the above points are fulfilled. Both COI and (Hymenoptera, Pteromalidae) and found the likely ND1 show identical haplotypes between the species co-transmission of Wolbachia and mtDNA from and the Swiss and Swedish populations of D. deletus. Nasonia giraulti Darling to Nasonia oneida There is strong evidence in the nuclear ITS2 marker Raychoudhury & Desjardins. Unusually large that D. deletus is more closely related to the North intraspecific mtDNA variation that coincides with American D. bradleyi; the sharing of an mtDNA hap- infections by different Wolbachia strains has been lotype in D. deletus and D. flixi through incomplete found in several studies (Ballard, Chernoff & James, lineage sorting is thus highly unlikely. The single 2002; Marshall, 2004; Riegler et al., 2005; Charlat Wolbachia infections in both species are of the same et al., 2009; Atyame et al., 2011; Xiao et al., 2012; strain type and only differ by a single mutation in the Ritter et al., 2013) but, without a clear hybridization highly variable wsp gene, an observation in accor- scenario including the species of origin, it might be dance with transmission through hybridization. that the bacterial strains became associated with Finally, D. deletus and D. flixi have been collected in divergent mtDNA haplotypes long after a transfer the same Malaise traps in the Swiss Alps which sug- event (or incomplete lineage sorting). gests geographical and phenological opportunity for hybridization. The case is similar for the species CONCLUSIONS D. parvus and D. varicoxa that share identical COI sequences both in Switzerland and Sweden, whereas Reports of endosymbiont-mediated mtDNA introgres- D. parvus clusters with D. tibiatorius in the ITS2 sion are very rare, with to our knowledge only six tree. Their multiple Wolbachia infections appear to convincing cases currently found in the literature. It partly overlap as well (because both species carry is difficult to estimate the prevalence and thus multiple infections, the strains could not be fully importance of this phenomenon. A recent study by typed). Wolbachia-mediated mtDNA transfer is thus the Consortium for the Barcode of Life (Smith et al., the likely cause for the failure of barcoding in these 2012) addressed Wolbachia and barcoding but cases, but insufficient variability or incomplete lin- mainly focussed on whether the CoxA gene (the bac- eage sorting probably also played a role (e.g., in the terial counterpart of COI) of the endosymbionts had sister species D. albotibialis and D. pectoratorius, been amplified instead of the sequences of the host which do not harbour Wolbachia infections). insects in recent barcoding initiatives; no conclusive There are very few other convincing examples for data was provided about hybrid introgression. endosymbiont-mediated mtDNA introgression in the Recent estimates of the prevalence of Wolbachia literature. Ballard (2000) could best explain the dis- infections on the one hand (20–65% of all insect spe-

© 2016 The Linnean Society of London, Zoological Journal of the Linnean Society, 2016 14 S. KLOPFSTEIN ET AL. cies; Hilgenboeker et al., 2008; Werren & Windsor, Tettelin H, Werren JH. 2006. Multilocus sequence typing 2000) and of hybridization on the other (approxi- system for the endosymbiont Wolbachia pipientis. Applied mately 10% of animal species; Mallet, 2005) suggest and Environmental Microbiology 72: 7098–7110. that endosymbiont-mediated mtDNA introgression Ballard JWO. 2000. When one is not enough: introgression might not be as rare as previously assumed. The of mitochondrial DNA in Drosophila. Molecular Biology and scarcity of empirical studies could simply be the Evolution 17: 1126–1130. result of an inherent difficulty to confirm this sce- Ballard JWO, Rand DM. 2005. The population biology of nario; most studies that include Wolbachia bacteria mitochondrial DNA and its phylogenetic implications. are conducted within species or they are part of a Annual Review of Ecology, Evolution, and Systematics 36: 621–642. broad screening study that does not assess mtDNA Ballard JWO, Chernoff B, James AC. 2002. Divergence of patterns in the hosts. Financial considerations and mitochondrial DNA is not corroborated by nuclear DNA, favourable reports on the success of DNA barcoding morphology, or behavior in Drosophila simulans. Evolution have led to many biodiversity studies including only 56: 527–545. a single mtDNA marker and consequently not Baur H, Leuenberger C. 2011. Analysis of ratiois in multi- detecting any mito-nuclear discordances. Decreasing variate morphometry. Systematic Biology 60: 813–825. sequencing costs, the establishment of additional Baur H, Kranz-Baltensperger Y, Cruaud A, Rasplus J- nuclear markers with sufficient variability, and the Y, Timokhov AV, Gokhman VE. 2014. Morphometric broad realization of the limitation of a single-marker analysis and taxonomic revision of Anisopteromalus species delimitation system will improve this situa- Ruschka (Hymenoptera: Chalcidoidea: Pteromalidae) - an tion and lead to a more precise assessment of the integrative approach. Systematic Entomology 39: 691–709. prevalence of this phenomenon. Belshaw R, Quicke DLJ. 1997. A molecular phylogeny of the Aphidiinae (Hymenoptera: Braconidae). Molecular Phy- logenetics and Evolution 7: 281–293. ACKNOWLEDGEMENTS Bergsten J, Bilton DT, Fujisawa T, Elliott M, Mon- We would like to thank the Swedish Malaise Trap aghan MT, Balke M, Hendrich L, Geijer J, Herrmann Project (Oland,€ Sweden), Victoria Spinas (Alp Flix, J, Foster GN, Ribera I, Nilsson AN, Barraclough TG, Switzerland), and Fridli Marti (Obersand, Switzer- Vogler AP. 2012. The effect of geographical scale of sam- land) for providing specimens. Madelaine Geiger and pling on DNA barcoding. Systematic Biology 61: 851–869. Salome Steiner performed the morphometric mea- Berthier P, Excoffier L, Ruedi M. 2006. Recurrent surements. This research received support from the replacement of mtDNA and cryptic hybridization between Swiss National Science Foundation (PA00P3_145377) two sibling bat species Myotis myotis and Myotis blythii. Proceedings of the Royal Society of London Series B-Biologi- and the SYNTHESYS Project (http://www.synthesys. cal Sciences 273: 3101–3109. info) (grant SE-TAF-663), which is financed by Euro- Beukeboom LW, Pijnacker LP. 2000. Automictic partheno- pean Community Research Infrastructure Action genesis in the parasitoid Venturia canescens (Hymenoptera: under the FP6 ‘Structuring the European Research Ichneumonidae) revisited. Genome 43: 939–944. Area’ Programme. Blaxter ML. 2004. The promise of a DNA taxonomy. Philo- sophical Transactions of the Royal Society of London Series REFERENCES B-Biological Sciences 359: 669–679. Braig HR, Zhou W, Dobson SL, O’Neill SL. 1998. Cloning Arthofer W, Riegler M, Schuler H, Schneider D, Moder and characterization of a gene encoding the major surface K, Miller WJ, Stauffer C. 2011. Allele intersection analy- protein of the bacterial endosymbiont Wolbachia pipientis. sis: a novel tool for multi locus sequence assignment in Journal of Bacteriology 180: 2373–2378. multiply infected hosts. PLoS ONE 6: e22198. Brown WM, Pragera EM, Wang A, Wilson AC. 1982. Atyame CM, Delsuc F, Pasteur N, Weill M, Duron O. Mitochondrial DNA sequences in primates: tempo and mode 2011. Diversification of Wolbachia endosymbiont in the of evolution. Journal of Molecular Evolution 18: 225–239. Culex pipiens mosquito. Molecular Biology and Evolution Butcher BA, Smith MA, Sharkey MJ, Quicke DLJ. 2012. 28: 2761–2772. A turbo-taxonomic study of Thai Aleiodes (Aleiodes) and Aubert J, Solignac M. 1990. Experimental evidence for Aleiodes (Arcaleiodes) (Hymenoptera: Braconidae: Rogadi- mitochondrial DNA introgression between Drosophila spe- nae) based largely on COI barcoded specimens, with rapid cies. Evolution 44: 1272–1282. descriptions of 179 new species. Zootaxa 3457: 1–232. Bachtrog D, Thornton K, Clark A, Andolfatto P. 2006. Charlat S, Duplouy A, Hornett EA, Dyson EA, Davies Extensive introgression of mitochondrial DNA relative to N, Roderick GK, Wedell N, Hurst GDD. 2009. The joint nuclear genes in the Drosophila yakuba species group. Evo- evolutionary histories of Wolbachia and mitochondria in lution 60: 292–302. Hypolimnas bolina. BMC Evolutionary Biology 9: 64. Baldo L, Hotopp JCD, Jolley KA, Bordenstein SR, Collins RA, Cruickshank RH. 2012. The seven deadly sins Biber SA, Choudhury RR, Hayashi C, Maiden MC, of DNA barcoding. Molecular Ecology Resources 13: 969–975.

Cook JM, Butcher RDJ. 1999. The transmission and ings of the Royal Society of London Series B-Biological effects of Wolbachia bacteria in parasitoids. Researches on Sciences 270: 313–321. Population Ecology 41: 15–28. Hebert PDN, Ratnasingham S, deWaard JR. 2003b. Derycke S, Fonseca G, Vierstraete A, Vanfleteren J, Barcoding animal life: cytochrome c oxidase subunit 1 Vincx M, Moens T. 2008. Disentangling taxonomy within divergences among closely related species. Proceedings of the Rhabditis (Pellioditis) marina (Nematoda, Rhabditidae) the Royal Society of London Series B-Biological Sciences species complex using molecular and morhological tools. 270: S96–S99. Zoological Journal of the Linnean Society 152: 1–15. Hebert PDN, Stoeckle MY, Zemlak TS, Francis CM. Duron O, Bouchon D, Boutin S, Bellamy L, Zhou L, 2004. Identification of birds through DNA barcodes. PLoS Engelstadter€ J, Hurst GDD. 2008. The diversity of repro- Biology 2: 1657–1663. ductive parasites among arthropods: Wolbachia do not walk Hilgenboeker K, Hammerstein P, Schlattmann P, Tel- alone. BMC Biology 6: 27. schow A, Werren JH. 2008. How many species are Edgar RC. 2004. MUSCLE: multiple sequence alignment infected with Wolbachia? - a statistical analysis of current with high accuracy and high throughput. Nucleic Acids data. FEMS Microbiology Letters 281: 215–220. Research 32: 1792–1797. Huelsenbeck JP, Larget B, Alfaro ME. 2004. Bayesian Edwards SV. 2009. Is a new and general theory of molecu- phylogenetic model selection using reversible-jump Markov lar systematics emerging? Evolution 63: 1–19. chain Monte Carlo. Molecular Biology and Evolution 21: Edwards SV, Bensch S. 2009. Looking forwards or looking 1123–1133. backwards in avian phylogeography? A comment on Zink Hurst GDD, Jiggins FM. 2005. Problems with mitochon- and Barrowclough 2008. Molecular Ecology 18: 2930–2933. drial DNA as a marker in population, phylogeographic and Feder JL, Forbes AA. 2010. Sequential speciation and the phylogenetic studies: the effects of inherited symbionts. diversity of parasitic insects. Ecological Entomology 35: 67– Proceedings of the Royal Society of London Series B-Biologi- 76. cal Sciences 272: 1525–1534. Folmer O, Black M, Hoeh W, Lutz R, Vrijenhoek R. Jiggins FM. 2003. Male-killing Wolbachia and mitochon- 1994. DNA primers for amplification of mitochondrial drial DNA: selective sweeps, hybrid introgression and para- cytochrome c oxidase subunit I from diverse metazoan site population dynamics. Genetics 164: 5–12. invertebrates. Molecular marine biology and biotechnology Johnstone RA, Hurst GDD. 1996. Maternally inherited 3: 294–299. male-killing microorganisms may confound interpretation Funk DJ, Omland KE. 2003. Species-level paraphyly and of mitochondrial DNA variability. Biological Journal of the polyphyly: frequency, causes, and consequences, with Linnean Society 58: 453–470. insights from animal mitochondrial DNA. Annual Review of Kass RE, Raftery AE. 1995. Bayes factors. Journal of Ecology, Evolution, and Systematics 34: 397–423. American Statistical Association 90: 773–795. Galtier N, Nabholz B, Glemin S, Hurst GDD. 2009. Klingenberg CP. 1998. Heterochrony and allometry: the Mitochondrial DNA as a marker of molecular diversity: a analysis of evolutionary change in ontogeny. Biological reappraisal. Molecular Ecology 18: 4541–4550. Reviews 73: 79–123. Gomez A, Wright PJ, Lunt DH, Cancino JM, Carvalho Klopfstein S. 2014. Revision of the Western Palaearctic GR, Hughes RN. 2007. Mating trials validate the use of Diplazontinae (Hymenoptera, Ichneumonidae). Zootaxa DNA barcoding to reveal cryptic speciation of a marine bry- 3801: 1–143. ozoan taxon. Proceedings of the Royal Society of London Klopfstein S, Ronquist F. 2013. Convergent intron gains in Series B-Biological Sciences 274: 199–207. hymenopteran elongation factor-1a. Molecular Phylogenetics Gompert Z, Forister ML, Fordyce JA, Nice CC. 2008. and Evolution 67: 266–276. Widespread mito-nuclear discordance with evidence for Klopfstein S, Kropf C, Quicke DLJ. 2010. An evaluation introgressive hybridization and selective sweeps in Lycaei- of phylogenetic informativeness profiles and the molecular des. Molecular Ecology 17: 5231–5244. phylogeny of Diplazontinae (Hymenoptera, Ichneumonidae). Hajibabaei M, Janzen DH, Burns JM, Hallwachs W, Systematic Biology 59: 226–241. Hebert PDN. 2006. DNA barcodes distinguish species of Klopfstein S, Quicke DLJ, Kropf C, Frick H. 2011. tropical Lepidoptera. Proceedings of the National Academy Molecular and morphological phylogeny of Diplazontinae of Sciences of the United States of America 103: 968–971. (Hymenoptera, Ichneumonidae). Zoologica Scripta 40: 379– Hartig G, Peters RS, Borner J, Etzbauer C, Misof B, 402. Niehuis O. 2012. Oligonucleotide primers for targeted Knowles LL, Carstens BC. 2007. Delimiting species with- amplification of single-copy nuclear genes in Apocritan out monophyletic gene trees. Systematic Biology 56: 887– Hymenoptera. PLoS ONE 7: e39826. 895. Heath BD, Butcher RDJ, Whitfield WGF, Hubbard SF. Konig€ K, Krimmer E, Brose S, Ganter C, Buschluter€ I, 1999. Horizontal transfer of Wolbachia between phylogenet- Konig€ C, Klopfstein S, Wendt I, Baur H, Krogmann L, ically distant insect species by a naturally occurring mecha- Steidle JLM. 2015. Does early learning drive ecological nism. Current Biology 9: 313–316. divergence during speciation processes in parasitoid wasps? Hebert PDN, Cywinska A, Ball SL, deWaard JR. 2003a. Proceedings of the Royal Society of London Series B-Biologi- Biological identifications through DNA barcodes. Proceed- cal Sciences 282: 20141850.

Lin C-P, Danforth BN. 2004. How do insect nuclear and taxonomy of undescribed insects. Systematic Biology 55: mitochondrial gene substitution patterns differ? Insights 595–609. from Bayesian analyses of combined datasets. Molecular Prufer K, Munch K, Hellmann I, Akagi K, Miller JR, Phylogenetics and Evolution 30: 686–702. Walenz B, Koren S, Sutton G, Kodira C, Winer R, Maddison WP, FitzJohn RG. 2015. The unsolved challenge Knight JR, Mullikin JC, Meader SJ, Ponting CP, Lun- to phylogenetic correlation tests for categorical characters. ter G, Higashino S, Hobolth A, Dutheil J, Karakoc E, Systematic Biology 64: 127–136. Alkan C, Sajjadian S, Catacchio CR, Ventura M, Mar- Mallet J. 2005. Hybridization as an invation of the genome. ques-Bonet T, Eichler EE, Andre C, Atencia R, Trends in Ecology & Evolution 20: 229–237. Mugisha L, Junhold J, Patterson N, Siebauer M, Mardulyn P, Whitfield JB. 1999. Phylogenetic signal in Good JM, Fischer A, Ptak SE, Lachmann M, Symer the COI, 16S, and 28S genes for inferring relationships DE, Mailund T, Schierup MH, Andres AM, Kelso J, among genera of Microgastrinae (Hymenoptera; Bra- Paabo S. 2012. The bonobo genome compared with the conidae): evidence of a high diversification rate in this chimpanzee and human genomes. Nature 486: 527–531. group of parasitoids. Molecular Phylogenetics and Evolution Quicke DLJ. 2012. We know too little about parasitoid wasp 12: 282–294. distributions to draw any conclusions about latitudinal Marshall JL. 2004. The Allonemobius-Wolbachia host-endo- trends in species richness, body size and biology. PLoS symbiont system: evidence for rapid speciation and against ONE 7: e32101. reproductive isolation driven by cytoplasmic incompatibil- Quicke DLJ, Mori M, Zaldivar-Riveron A, Laurenne ity. Evolution 58: 2409–2425. NM, Shaw MR. 2006. Suspended mummies in Aleiodes Meier R, Shiyang K, Vaidya G, Ng PKL. 2006. DNA bar- species (Hymenoptera: Braconidae: Rogadinae) with coding and taxonomy in Diptera: a tale of high intraspecific descriptions of six new species from western Uganda based variability and low identification success. Systematic Biol- largely on DNA sequence data. Journal of Natural History ogy 55: 715–728. 40: 2663–2680. Meier R, Zhang G, Ali F. 2008. The use of mean instead of R Core Team. 2014. R: A language and environment for sta- smallest interspecific distance exaggerates the size of the tistical computing. R Foundation for Statistical Computing, “barcoding gap” and leads to misidentification. Systematic Vienna, Austria. Biology 57: 809–813. Raychoudhury R, Baldo L, Oliveira DCSG, Werren JH. Mindell DP, Thacker CE. 1996. Rates of molecular evolu- 2009. Modes of acquisition of Wolbachia: horizontal trans- tion: phylogenetic issues and applications. Annual Review fer, hybrid introgression, and codivergence in the Nasonia of Ecology and Systematics 27: 279–303. species complex. Evolution 63: 165–183. Mueller RL. 2006. Evolutionary rates, divergence dates, and Riegler M, Sidhu M, Miller WJ, O’Neill SL. 2005. Evi- the performance of mitochondrial genes in Bayesian phylo- dence for a global Wolbachia replacement in Drosophila genetic analysis. Systematic Biology 55: 289–300. melanogaster. Current Biology 15: 1428–1433. Narita S, Nomura M, Kato Y, Fukatsu T. 2006. Genetic Ritter S, Michalski SG, Settele J, Wiemers M, Fric ZF, structure of sibling butterfly species affected by Wolbachia Sielezniew M, Sasic M, Rozier Y, Durka W. 2013. Wol- infection sweep: evolutionary and biogeographical implica- bachia infections mimic cryptic speciation in two parasitic tions. Molecular Ecology 15: 1095–1108. butterfly species, Phengaris teleius and P. nausithous (Lepi- Nicholls JA, Challis RJ, Mutun S, Stone GN. 2012. Mito- dopters: Lycaenidae). PLoS ONE 8: e78107. chondrial barcodes are diagnostic of shared refugia but not Ronquist F, Teslenko M, Van der Mark P, Ayres DL, species in hybridiying oak gallwasps. Molecular Ecology 21: Darling A, Hohna€ S, Larget B, Liu L, Suchard MA, 4051–4062. Huelsenbeck JP. 2012. MrBayes 3.2: efficient Bayesian Pagel M, Meade A. 2006. Bayesian analysis of correlated phylogenetic inference and model choice across a large evolution of discrete characters by reversible-jump Markov model space. Systematic Biology 61: 539–542. chain Monte Carlo. American Naturalist 167: 808–825. Rosenberg NA, Tao R. 2008. Discordance of species trees Paquin P, Hedin M. 2004. The power and perils of ‘molecu- with their most likely gene trees: the case of five taxa. Sys- lar taxonomy’: a case study of eyeless and endangered tematic Biology 57: 131–140. Cicurina (Araneae: Dictynidae) from Texas caves. Molecu- Rubinoff D, Cameron S, Will K. 2006. A genomic perspec- lar Ecology 13: 3239–3255. tive on the shortcomings of mitochondrial DNA for ‘‘barcod- Petit RJ, Excoffier L. 2009. Gene flow and species delimita- ing’’ identification. Journal of Heredity 97: 581–594. tion. Trends in Ecology & Evolution 24: 386–393. Schmidt S, Schmid-Egger C, Moriniere J, Haszprunar Pimm SL, Jenkins CN, Abell R, Brooks TM, Gittleman G, Hebert PDN. 2015. DNA barcoding largely supports JL, Joppa LN, Raven PH, Roberts CM, Sexton JO. 250 years of classical taxonomy: identications for Central 2014. The biodiversity of species and their rates of European bees (Hymenoptera, Apoidea partim). Molecular extinction, distribution, and protection. Science 344: Ecology Resources 15: 985–1000. 1246752. Schweizer M, Hertwig S, Seehausen O. 2014. Diversity Pons J, Barraclough TG, Gomez-Zurita J, Cardoso A, versus disparity and the role of ecological opportunity in a Duran DP, Hazell S, Kamoun S, Sumlin WD, Vogler continental bird radiation. Journal of Biogeography 41: AP. 2006. Sequence-based species delimitation for the DNA 1301–1312.

Simoes~ PM, Mialdea G, Reiss D, Sagot M-F, Charlat S. Turelli M, Hoffmann AA, McKechnie SW. 1992. Dynam- 2011. Wolbachia detection: an assessment of standard PCR ics of cytoplasmic incompatibility and mtDNA variation in protocols. Molecular Ecology Resources 11: 567–572. natural Drosophila simulans populations. Genetics 132: Simon C, Buckley TR, Frati F, Stewart JB, Beckenbach 713–723. AT. 2006. Incorporating molecular evolution into phyloge- Vavre F, Fleury F, Lepetit D, Fouillet P, Bouletreau M. netic analysis, and a new compilation of conserved poly- 1999. Phylogenetic evidence for horizontal transmission of merase chain reaction primers for animal mitochondrial Wolbachia in host-parasitoid associations. Molecular Biol- DNA. Annual Review of Ecology, Evolution, and Systemat- ogy and Evolution 16: 1711–1723. ics 37: 545–579. Vogler AP, Monaghan MT. 2007. Recent advances in DNA Smith MA, Rodriguez JJ, Whitfield JB, Deans AR, Jan- taxonomy. Journal of Zoological Systematics and Evolution- zen DH, Hallwachs W, Hebert PDN. 2008. Extreme ary Research 45: 1–10. diversity of tropical parasitoid wasps exposed by iterative Werren JH, Windsor DM. 2000. Wolbachia infection fre- integration of natural history, DNA barcoding, morphology, quencies in insects: evidence of a global equilibrium? Pro- and collections. Proceedings of the National Academy of ceedings of the Royal Society of London Series B-Biological Sciences of the United States of America 105: 12359–12364. Sciences 267: 1277–1285. Smith MA, Bertrand C, Crosby K, Eveleigh ES, Fernan- Werren JH, Baldo L, Clark ME. 2008. Wolbachia: master dez-Triana J, Fisher BL, Gibbs J, Hajibabaei M, Hall- manipulators of invertebrate biology. Nature Reviews - wachs W, Hind K, Hrcek J, Huang D-W, Janda M, Microbiology 6: 741–751. Janzen DH, Li Y, Miller SE, Packer L, Quicke DLJ, Whitworth TL, Dawson RD, Magalon H, Baudry E. Ratnasingham S, Rodriguez JJ, Rougerie R, Shaw 2007. DNA barcoding cannot reliably identify species of the MR, Sheffield C, Stahlhut JK, Steinke D, Whitfield blowfly genus Protocalliphora (Diptera: Calliphoridae). Pro- JB, Wood M, Zhou X. 2012. Wolbachia and DNA barcod- ceedings of the Royal Society of London Series B-Biological ing insects: patterns, potential, and problems. PLoS ONE 7: Sciences 274: 1731–1739. e36514. Wickham H. 2009. ggplot2 - elegant graphics for data analy- Stamatakis A. 2014. RAxML version 8: a tool for phyloge- sis. Dordrecht: Springer. netic analysis and post-analysis of large phylogenies. Bioin- Xiao J-H, Wang N-X, Murphy RW, Cook JM, Jia L-Y, formatics 30: 1312–1313. Huang D-W. 2012. Wolbachia infection and dramatic Stigenberg J, Ronquist F. 2011. Revision of the Western intraspecific mitochondrial DNA divergence in a fig wasp. Palearctic Meteorini (Hymenoptera, Braconidae), with a Evolution 66: 1907–1916. molecular characterization of hidden Fennoscandian species Yang Z, Rannala B. 2010. Bayesian species delimitation diversity. Zootaxa 3084: 1–95. using multilocus sequence data. Proceedings of the National Tamura K, Stecher G, Peterson D, Filipsky A, Kumar S. Academy of Sciences of the United States of America 107: 2013. MEGA6: molecular evolutionary genetics analysis 9264–9269. version 6.0. Molecular Biology and Evolution 30: 2725– Zhang J, Kapli P, Pavlidis P, Stamatakis A. 2013. A gen- 2729. eral species delimitation method with applications to phylo- Tautz D, Arctander P, Minelli A, Thomas RH, Vogler genetic placements. Bioinformatics 29: 2869–2876. AP. 2002. DNA points the way ahead in taxonomy. Nature Zink RM, Barrowclough GF. 2008. Mitochondrial DNA 418: 479. under siege in avian phylogeography. Molecular Ecology 17: Townes HK. 1969. The genera of Ichneumonidae, Part 1. 2107–2121. Memoirs of the American Entomological Institute 11: 1–300.

SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article at the publisher’s web- site: Figure S1. Scatterplot of the ﬁrst shape principle component versus isometric size in selected species pairs of the genus Diplazon. Figure S2. Bayesian consensus trees of the concatenate dataset of ﬁve genes, and single-gene trees for the nuclear 28S rDNA and EF1a genes and the mitochondrial NADH1. Table S1. List of specimens and Genbank accession numbers. Table S2. Description of characters used in morphometrics of Diplazon. Table S3. Measurement data from morphometric analysis.